Monotone return to zero needs changes above 0.10 in 467 of 607 selected Allen recovery curves

We tested a necessary shape constraint for one-state recovery after a stimulus train. We used first-response recovery summaries at four pauses in Allen Synaptic Physiology r2.1. Each pause had at least five trials. Of 607 detected cortical connections, 467 needed a uniform change above 0.10 normalized units to permit a monotone return toward zero. The counts were 168 of 217 for mouse excitation, 258 of 329 for mouse inhibition, 38 of 57 for human excitation, and 3 of 4 for human inhibition. Without the zero constraint, 340 curves still needed a change above 0.10 to permit either monotone direction. An exact formula agreed with linear optimization in 1229 observed and synthetic cases. The tolerance is an assumption, not a measured error bound. Noise, trial selection, protocol drift, and condition-specific baselines prevent a claim of biological model rejection. The package contains code, a proof, aggregate evidence, a figure, source hashes, and a retrieval method. External record-level data remain outside the package.

Allen Institute · synaptic recovery · shape constraint · model compatibility · measurement tolerance